linux-kernel - Re: [PATCH v2 2/7] media: chips-media: wave5: Improve performance of decoder

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <8c6316d2d00287e96bc2ad5d27c8de37f990b7e5.camel@collabora.com>
Date: Thu, 05 Jun 2025 09:28:37 -0400
From: Nicolas Dufresne <nicolas.dufresne@...labora.com>
To: "jackson.lee" <jackson.lee@...psnmedia.com>, "mchehab@...nel.org"	
 <mchehab@...nel.org>, "hverkuil-cisco@...all.nl"
 <hverkuil-cisco@...all.nl>,  "bob.beckett@...labora.com"	
 <bob.beckett@...labora.com>
Cc: "linux-media@...r.kernel.org" <linux-media@...r.kernel.org>, 
 "linux-kernel@...r.kernel.org"	 <linux-kernel@...r.kernel.org>,
 "lafley.kim" <lafley.kim@...psnmedia.com>,  "b-brnich@...com"	
 <b-brnich@...com>, "hverkuil@...all.nl" <hverkuil@...all.nl>, Nas Chung	
 <nas.chung@...psnmedia.com>
Subject: Re: [PATCH v2 2/7] media: chips-media: wave5: Improve performance
 of decoder

Hi,

Le jeudi 05 juin 2025 à 04:50 +0000, jackson.lee a écrit :
> > So my question is, is there a way to know, at CMD_STOP call, that the HW
> > has gone idle, and that no more events will allow handling the EOS case?
> > 
> > Nicolas
> 
> 
> Thanks for your reply.	
> 
> Now there is only one thing to know if there is more events or not to handle the EOS case.
> It is that driver sends DEC_PIC command to VPU continuously until display index is -2(it means EOS) from VPU.
> VPU should trigger interrupts to get display index from the finish_decode function.
> So we have to run device_run to send DEC_PIC command.

What don't want to see is a loop where we do:

	device_run()
		finish_job()
			device_run()
				finish_job()
					....

What I see now, is that we simply bang on the trigger until it completes, which
is very wasteful in power and CPU time. In your next version, make sure to
find a mitigation to that active loop, and document it please.

Nicolas